Search

School districts have historically operated summer programs to give students extra learning and enrichment opportunities to promote positive academic and behavioral outcomes. Despite the potential benefits and policymakers' endorsement of summer programming, recent research suggests that summer programs have a limited impact on post-pandemic academic recovery, with modest gains in math test scores and no impact in reading. This may be partly due to the challenges districts face in implementing academic recovery programs at the scale needed to impact student outcomes. This brief highlights the efficacy of summer programs for literacy, math and social-emotional learning (SEL) outcomes, showing that well-designed summer programs can meaningfully boost student achievement. The report also provides recommendations for district leaders to use as a framework for planning and implementing effective summer programming.

The recent rise in increasingly sophisticated cyber-attacks raises the need for robust and resilient autonomous cyber-defence (ACD) agents. Given the variety of cyber-attack tactics, techniques and procedures (TTPs) employed, learning approaches that can return generalisable policies are desirable. Meanwhile, the assurance of ACD agents remains an open challenge. We address both challenges via an empirical game-theoretic analysis of deep reinforcement learning (DRL) approaches for ACD using the principled double oracle (DO) algorithm. This algorithm relies on adversaries iteratively learning (approximate) best responses against each others' policies; a computationally expensive endeavour for autonomous cyber operations agents. In this work we introduce and evaluate a theoretically-sound, potential-based reward shaping approach to expedite this process. In addition, given the increasing number of open-source ACD-DRL approaches, we extend the DO formulation to allow for multiple response oracles (MRO), providing a framework for a holistic evaluation of ACD approaches., Comment: 21 pages, 17 figures, 10 tables

The flow of information within many-body systems is a fundamental feature of physical interaction. Given an underlying classical physics model for the interaction between a particle and its environment, we give meaning to and quantify the information passed between them over time. We show that the maximum information exchange rate is proportional to the ratio of inter-particle energy flow and initial particle energy -- a sort of signal-to-noise ratio. In addition, a single time-point (as opposed to trajectory) observability relation emerges.

Exchanging the positions of two non-Abelian anyons transforms between many-body wavefunctions within a degenerate ground-state manifold. This behavior is fundamentally distinct from fermions, bosons and Abelian anyons. Recently, quantum dot-superconductor arrays have emerged as a promising platform for creating topological Kitaev chains that can host non-Abelian Majorana zero modes. In this work, we propose a minimal braiding setup in a linear array of quantum dots consisting of two minimal Kitaev chains coupled through an ancillary, normal quantum dot. We focus on the physical effects that are peculiar to quantum dot devices, such as interdot Coulomb repulsion and residual single electron tunneling. We find that the errors caused by either of these effects can be efficiently mitigated by optimal control of the ancillary quantum dot that mediates the exchange of the non-Abelian anyons. Moreover, we propose experimentally accessible methods to find this optimal operating regime and predict signatures of a successful Majorana braiding experiment., Comment: 17 pages, 8 figures

Sunny is a Julia package designed to serve the needs of the quantum magnetism community. It supports the specification of a very broad class of spin models and a diverse suite of numerical solvers. These include powerful methods for simulating spin dynamics both in and out of equilibrium. Uniquely, it features a broad generalization of classical and semiclassical approaches to SU(N) coherent states, which is useful for studying systems exhibiting strong spin-orbit coupling or local entanglement effects. Sunny also offers a well-developed framework for calculating the dynamical spin structure factor, enabling direct comparison with scattering experiments. Ease of use is a priority, with tools for symmetry-guided modeling and interactive visualization.

We present PDLP, a practical first-order method for linear programming (LP) designed to solve large-scale LP problems. PDLP is based on the primal-dual hybrid gradient (PDHG) method applied to the minimax formulation of LP. PDLP incorporates several enhancements to PDHG, including diagonal preconditioning, presolving, adaptive step sizes, adaptive restarting, and feasibility polishing. Our algorithm is implemented in C++, available in Google's open-source OR-Tools library, and supports multithreading. To evaluate our method, we introduce a new collection of eleven large-scale LP problems with sizes ranging from 125 million to 6.3 billion nonzeros. PDLP solves eight of these instances to optimality gaps of 1\% (with primal and dual feasibility errors of less than $10^{-8}$) within six days on a single machine. We also compare PDLP with Gurobi barrier, primal simplex, and dual simplex implementations. Gurobi barrier solves only three instances, exceeding our 1TB RAM limit on the other eight. While primal and dual simplex are more memory-efficient than the barrier method, they are slower and solve only three instances within six days. Compared with the conference version of this work (in: Advances in Neural Information Processing Systems 34 (NeurIPS 2021)), the key new contributions are: (i) feasibility polishing, a technique that quickly finds solutions that are approximately optimal but almost exactly feasible (without which only three of the eleven problems can be solved); (ii) a multithreaded C++ implementation available in Google OR-Tools; and (iii) a new collection of large-scale LP problems. Note that the conference version should be referred to for comparisons with SCS and ablation studies, which we do not repeat in this paper.

Optimizing charged-particle track reconstruction algorithms is crucial for efficient event reconstruction in Large Hadron Collider (LHC) experiments due to their significant computational demands. Existing track reconstruction algorithms have been adapted to run on massively parallel coprocessors, such as graphics processing units (GPUs), to reduce processing time. Nevertheless, challenges remain in fully harnessing the computational capacity of coprocessors in a scalable and non-disruptive manner. This paper proposes an inference-as-a-service approach for particle tracking in high energy physics experiments. To evaluate the efficacy of this approach, two distinct tracking algorithms are tested: Patatrack, a rule-based algorithm, and Exa$.$TrkX, a machine learning-based algorithm. The as-a-service implementations show enhanced GPU utilization and can process requests from multiple CPU cores concurrently without increasing per-request latency. The impact of data transfer is minimal and insignificant compared to running on local coprocessors. This approach greatly improves the computational efficiency of charged particle tracking, providing a solution to the computing challenges anticipated in the High-Luminosity LHC era., Comment: 19 pages, 8 figures, submitted to JINST

We provided comments on the article Orbits of particles with magnetic dipole moment around magnetized Schwarzschild black holes: Applications to S2 star orbit by Uktamjon Uktamov, Mohsen Fathi, Javlon Rayimbaev, and Ahmadjon Abdujabbarov from arXiv:2406.03371v2. We derived the Hamilton-Jacobi equation used in the article from the Lagrangian utilized and found inconsistencies in the use of the interaction term which describes the magnetic dipole moment of the particle. Consequently, this results in incorrect equations of motion for the particle., Comment: 5 pages; comment on arXiv:2406.03371v2

We demonstrate trapping of individual rubidium (Rb) and cesium (Cs) atoms in an interleaved array of bright tweezers and dark bottle-beam traps, using a microfabricated optical element illuminated by a single laser beam. This trapping approach is simple and robust compared to common approaches that use active optoelectronic components and multiple laser wavelengths, and can be easily adapted for other atomic species.

The o1 model series is trained with large-scale reinforcement learning to reason using chain of thought. These advanced reasoning capabilities provide new avenues for improving the safety and robustness of our models. In particular, our models can reason about our safety policies in context when responding to potentially unsafe prompts, through deliberative alignment. This leads to state-of-the-art performance on certain benchmarks for risks such as generating illicit advice, choosing stereotyped responses, and succumbing to known jailbreaks. Training models to incorporate a chain of thought before answering has the potential to unlock substantial benefits, while also increasing potential risks that stem from heightened intelligence. Our results underscore the need for building robust alignment methods, extensively stress-testing their efficacy, and maintaining meticulous risk management protocols. This report outlines the safety work carried out for the OpenAI o1 and OpenAI o1-mini models, including safety evaluations, external red teaming, and Preparedness Framework evaluations.

How are AI assistants being used in the real world? While model providers in theory have a window into this impact via their users' data, both privacy concerns and practical challenges have made analyzing this data difficult. To address these issues, we present Clio (Claude insights and observations), a privacy-preserving platform that uses AI assistants themselves to analyze and surface aggregated usage patterns across millions of conversations, without the need for human reviewers to read raw conversations. We validate this can be done with a high degree of accuracy and privacy by conducting extensive evaluations. We demonstrate Clio's usefulness in two broad ways. First, we share insights about how models are being used in the real world from one million Claude.ai Free and Pro conversations, ranging from providing advice on hairstyles to providing guidance on Git operations and concepts. We also identify the most common high-level use cases on Claude.ai (coding, writing, and research tasks) as well as patterns that differ across languages (e.g., conversations in Japanese discuss elder care and aging populations at higher-than-typical rates). Second, we use Clio to make our systems safer by identifying coordinated attempts to abuse our systems, monitoring for unknown unknowns during critical periods like launches of new capabilities or major world events, and improving our existing monitoring systems. We also discuss the limitations of our approach, as well as risks and ethical concerns. By enabling analysis of real-world AI usage, Clio provides a scalable platform for empirically grounded AI safety and governance.

Pretraining large language models is a complex endeavor influenced by multiple factors, including model architecture, data quality, training continuity, and hardware constraints. In this paper, we share insights gained from the experience of training DMaS-LLaMa-Lite, a fully open source, 1.7-billion-parameter, LLaMa-based model, on approximately 20 billion tokens of carefully curated data. We chronicle the full training trajectory, documenting how evolving validation loss levels and downstream benchmarks reflect transitions from incoherent text to fluent, contextually grounded output. Beyond pretraining, we extend our analysis to include a post-training phase focused on instruction tuning, where the model was refined to produce more contextually appropriate, user-aligned responses. We highlight practical considerations such as the importance of restoring optimizer states when resuming from checkpoints, and the impact of hardware changes on training stability and throughput. While qualitative evaluation provides an intuitive understanding of model improvements, our analysis extends to various performance benchmarks, demonstrating how high-quality data and thoughtful scaling enable competitive results with significantly fewer training tokens. By detailing these experiences and offering training logs, checkpoints, and sample outputs, we aim to guide future researchers and practitioners in refining their pretraining strategies. The training script is available on Github at https://github.com/McGill-DMaS/DMaS-LLaMa-Lite-Training-Code. The model checkpoints are available on Huggingface at https://huggingface.co/collections/McGill-DMaS/dmas-llama-lite-6761d97ba903f82341954ceb.

We introduce SimAvatar, a framework designed to generate simulation-ready clothed 3D human avatars from a text prompt. Current text-driven human avatar generation methods either model hair, clothing, and the human body using a unified geometry or produce hair and garments that are not easily adaptable for simulation within existing simulation pipelines. The primary challenge lies in representing the hair and garment geometry in a way that allows leveraging established prior knowledge from foundational image diffusion models (e.g., Stable Diffusion) while being simulation-ready using either physics or neural simulators. To address this task, we propose a two-stage framework that combines the flexibility of 3D Gaussians with simulation-ready hair strands and garment meshes. Specifically, we first employ three text-conditioned 3D generative models to generate garment mesh, body shape and hair strands from the given text prompt. To leverage prior knowledge from foundational diffusion models, we attach 3D Gaussians to the body mesh, garment mesh, as well as hair strands and learn the avatar appearance through optimization. To drive the avatar given a pose sequence, we first apply physics simulators onto the garment meshes and hair strands. We then transfer the motion onto 3D Gaussians through carefully designed mechanisms for each body part. As a result, our synthesized avatars have vivid texture and realistic dynamic motion. To the best of our knowledge, our method is the first to produce highly realistic, fully simulation-ready 3D avatars, surpassing the capabilities of current approaches., Comment: Project website: https://nvlabs.github.io/SimAvatar/

In 1946, Garrett Birkhoff proved that the $n\times n$ doubly stochastic matrices comprise the convex hull of the $n\times n$ permutation matrices, which in turn make up the extreme points of this polytope. He proposed his problem 111, which asks whether there exists a topology on infinite matrices for which this applies to the closed convex hull of the $\mathbb{N}\times\mathbb{N}$ permutation matrices. As Isbell showed in 1955, this equality is not achieved in the line-sum norm. In this paper, we use the domain of operator theory, and its many topologies, to improve on his negative result by showing that Birkhoff's problem is not solved in any of these topologies. In Kendall's 1960 paper on this problem, he gave an answer to the affirmative, as well as a topology for which closed convex hull comprises the doubly substochastic matrices. We also show that Kendall's secondary theorem also applies for all the locally convex Hausdorff topologies finer than than Kendall's (namely that of entry-wise convergence) which make the continuous dual of the matrix space no larger than the predual of the von Neumann algebra containing them. We then show that this is a theoretical upper limit topologies with this closure property. We also discuss the exposed points of this hull for these several topologies. Moreover, we show that, in these topologies, the closed affine hull of these permutation matrices comprise all operators with real-entry matrix coefficients.

We articulate fundamental mismatches between technical methods for machine unlearning in Generative AI, and documented aspirations for broader impact that these methods could have for law and policy. These aspirations are both numerous and varied, motivated by issues that pertain to privacy, copyright, safety, and more. For example, unlearning is often invoked as a solution for removing the effects of targeted information from a generative-AI model's parameters, e.g., a particular individual's personal data or in-copyright expression of Spiderman that was included in the model's training data. Unlearning is also proposed as a way to prevent a model from generating targeted types of information in its outputs, e.g., generations that closely resemble a particular individual's data or reflect the concept of "Spiderman." Both of these goals--the targeted removal of information from a model and the targeted suppression of information from a model's outputs--present various technical and substantive challenges. We provide a framework for thinking rigorously about these challenges, which enables us to be clear about why unlearning is not a general-purpose solution for circumscribing generative-AI model behavior in service of broader positive impact. We aim for conceptual clarity and to encourage more thoughtful communication among machine learning (ML), law, and policy experts who seek to develop and apply technical methods for compliance with policy objectives., Comment: Presented at the 2nd Workshop on Generative AI and Law at ICML (July 2024)

Model-based reinforcement learning (MBRL) is a promising route to sample-efficient policy optimization. However, a known vulnerability of reconstruction-based MBRL consists of scenarios in which detailed aspects of the world are highly predictable, but irrelevant to learning a good policy. Such scenarios can lead the model to exhaust its capacity on meaningless content, at the cost of neglecting important environment dynamics. While existing approaches attempt to solve this problem, we highlight its continuing impact on leading MBRL methods -- including DreamerV3 and DreamerPro -- with a novel environment where background distractions are intricate, predictable, and useless for planning future actions. To address this challenge we develop a method for focusing the capacity of the world model through synergy of a pretrained segmentation model, a task-aware reconstruction loss, and adversarial learning. Our method outperforms a variety of other approaches designed to reduce the impact of distractors, and is an advance towards robust model-based reinforcement learning., Comment: Accepted at NeurIPS 2024

We present the MULTIMODAL UNIVERSE, a large-scale multimodal dataset of scientific astronomical data, compiled specifically to facilitate machine learning research. Overall, the MULTIMODAL UNIVERSE contains hundreds of millions of astronomical observations, constituting 100\,TB of multi-channel and hyper-spectral images, spectra, multivariate time series, as well as a wide variety of associated scientific measurements and "metadata". In addition, we include a range of benchmark tasks representative of standard practices for machine learning methods in astrophysics. This massive dataset will enable the development of large multi-modal models specifically targeted towards scientific applications. All codes used to compile the MULTIMODAL UNIVERSE and a description of how to access the data is available at https://github.com/MultimodalUniverse/MultimodalUniverse, Comment: Accepted at NeurIPS Datasets and Benchmarks track

In an accompanying publication, the MeerKAT Pulsar Timing Array (MPTA) collaboration reports tentative evidence for the presence of a stochastic gravitational-wave background, following observations of similar signals from the European and Indian Pulsar Timing Arrays, NANOGrav, the Parkes Pulsar Timing Array and the Chinese Pulsar Timing Array. If such a gravitational-wave background signal originates from a population of inspiraling supermassive black-hole binaries, the signal may be anisotropically distributed on the sky. In this Letter we evaluate the anisotropy of the MPTA signal using a spherical harmonic decomposition. We discuss complications arising from the covariance between pulsar pairs and regularisation of the Fisher matrix. Applying our method to the 4.5 yr dataset, we obtain two forms of sky maps for the three most sensitive MPTA frequency bins between 7 -21 nHz. Our "clean maps'' estimate the distribution of gravitational-wave strain power with minimal assumptions. Our radiometer maps answer the question: is there a statistically significant point source? We find a noteworthy hotspot in the 7 nHz clean map with a $p$-factor of $p=0.015$ (not including trial factors). Future observations are required to determine if this hotspot is of astrophysical origin.

Pulsar Timing Arrays search for nanohertz-frequency gravitational waves by regularly observing ensembles of millisecond pulsars over many years to look for correlated timing residuals. Recently the first evidence for a stochastic gravitational wave background has been presented by the major Arrays, with varying levels of significance ($\sim$2-4$\sigma$). In this paper we present the results of background searches with the MeerKAT Pulsar Timing Array. Although of limited duration (4.5 yr), the $\sim$ 250,000 arrival times with a median error of just $3 \mu$s on 83 pulsars make it very sensitive to spatial correlations. Detection of a gravitational wave background requires careful modelling of noise processes to ensure that any correlations represent a fit to the underlying background and not other misspecified processes. Under different assumptions about noise processes we can produce either what appear to be compelling Hellings-Downs correlations of high significance (3-3.4$\sigma$) with a spectrum close to that which is predicted, or surprisingly, under slightly different assumptions, ones that are insignificant. This appears to be related to the fact that many of the highest precision MeerKAT Pulsar Timing Array pulsars are in close proximity and dominate the detection statistics. The sky-averaged characteristic strain amplitude of the correlated signal in our most significant model is $h_{c, {\rm yr}} = 7.5^{+0.8}_{-0.9} \times 10^{-15}$ measured at a spectral index of $\alpha=-0.26$, decreasing to $h_{c, {\rm yr}} = 4.8^{+0.8}_{-0.9} \times 10^{-15}$ when assessed at the predicted $\alpha=-2/3$. These data will be valuable as the International Pulsar Timing Array project explores the significance of gravitational wave detections and their dependence on the assumed noise models.

Pulsar timing arrays are ensembles of regularly observed millisecond pulsars timed to high precision. Each pulsar in an array could be affected by a suite of noise processes, most of which are astrophysically motivated. Analysing them carefully can be used to understand these physical processes. However, the primary purpose of these experiments is to detect signals that are common to all pulsars, in particular signals associated with a stochastic gravitational wave background. To detect this, it is paramount to appropriately characterise other signals that may otherwise impact array sensitivity or cause a spurious detection. Here we describe the second data release and first detailed noise analysis of the pulsars in the MeerKAT Pulsar Timing Array, comprising high-cadence and high-precision observations of $83$ millisecond pulsars over $4.5$ years. We use this analysis to search for a common signal in the data, finding a process with an amplitude of $\log_{10}\mathrm{A_{CURN}} = -14.25^{+0.21}_{-0.36}$ and spectral index $\gamma_\mathrm{CURN} = 3.60^{+1.31}_{-0.89}$. Fixing the spectral index at the value predicted for a background produced by the inspiral of binary supermassive black holes, we measure the amplitude to be $\log_{10}\mathrm{A_{CURN}} = -14.28^{+0.21}_{-0.21}$ at a significance expressed as a Bayes factor of $\ln(\mathcal{B}) = 4.46$. Under both assumptions, the amplitude that we recover is larger than those reported by other PTA experiments. We use the results of this analysis to forecast our sensitivity to a gravitational wave background possessing the spectral properties of the common signal we have measured.

While the challenges and solutions for efficient execution of scalable vector ISAs on long-vector-length microarchitectures have been well established, not all of these solutions are suitable for short-vector-length implementations. This work proposes a novel microarchitecture for instruction sequencing in vector units with short architectural vector lengths. The proposed microarchitecture supports fine-granularity chaining, multi-issue out-of-order execution, zero dead-time, and run-ahead memory accesses with low area or complexity costs. We present the Saturn Vector Unit, a RTL implementation of a RVV vector unit. With our instruction scheduling mechanism, Saturn exhibits comparable or superior power, performance, and area characteristics compared to state-of-the-art long-vector and short-vector implementations.

GPT-4o is an autoregressive omni model that accepts as input any combination of text, audio, image, and video, and generates any combination of text, audio, and image outputs. It's trained end-to-end across text, vision, and audio, meaning all inputs and outputs are processed by the same neural network. GPT-4o can respond to audio inputs in as little as 232 milliseconds, with an average of 320 milliseconds, which is similar to human response time in conversation. It matches GPT-4 Turbo performance on text in English and code, with significant improvement on text in non-English languages, while also being much faster and 50\% cheaper in the API. GPT-4o is especially better at vision and audio understanding compared to existing models. In line with our commitment to building AI safely and consistent with our voluntary commitments to the White House, we are sharing the GPT-4o System Card, which includes our Preparedness Framework evaluations. In this System Card, we provide a detailed look at GPT-4o's capabilities, limitations, and safety evaluations across multiple categories, focusing on speech-to-speech while also evaluating text and image capabilities, and measures we've implemented to ensure the model is safe and aligned. We also include third-party assessments on dangerous capabilities, as well as discussion of potential societal impacts of GPT-4o's text and vision capabilities.

Machine learning based surrogate models offer researchers powerful tools for accelerating simulation-based workflows. However, as standard datasets in this space often cover small classes of physical behavior, it can be difficult to evaluate the efficacy of new approaches. To address this gap, we introduce the Well: a large-scale collection of datasets containing numerical simulations of a wide variety of spatiotemporal physical systems. The Well draws from domain experts and numerical software developers to provide 15TB of data across 16 datasets covering diverse domains such as biological systems, fluid dynamics, acoustic scattering, as well as magneto-hydrodynamic simulations of extra-galactic fluids or supernova explosions. These datasets can be used individually or as part of a broader benchmark suite. To facilitate usage of the Well, we provide a unified PyTorch interface for training and evaluating models. We demonstrate the function of this library by introducing example baselines that highlight the new challenges posed by the complex dynamics of the Well. The code and data is available at https://github.com/PolymathicAI/the_well., Comment: 38th Conference on Neural Information Processing Systems (NeurIPS 2024) Track on Datasets and Benchmarks

Multi-agent systems (MAS) utilizing multiple Large Language Model agents with Retrieval Augmented Generation and that can execute code locally may become beneficial in cosmological data analysis. Here, we illustrate a first small step towards AI-assisted analyses and a glimpse of the potential of MAS to automate and optimize scientific workflows in Cosmology. The system architecture of our example package, that builds upon the autogen/ag2 framework, can be applied to MAS in any area of quantitative scientific research. The particular task we apply our methods to is the cosmological parameter analysis of the Atacama Cosmology Telescope lensing power spectrum likelihood using Monte Carlo Markov Chains. Our work-in-progress code is open source and available at https://github.com/CMBAgents/cmbagent., Comment: Comments welcome!

We present improved approximation algorithms for some problems in the related areas of Flexible Graph Connectivity and Capacitated Network Design. In the $(p,q)$-Flexible Graph Connectivity problem, denoted $(p,q)$-FGC, the input is a graph $G(V, E)$ where $E$ is partitioned into safe and unsafe edges, and the goal is to find a minimum cost set of edges $F$ such that the subgraph $G'(V, F)$ remains $p$-edge connected upon removal of any $q$ unsafe edges from $F$. In the related Cap-$k$-ECSS problem, we are given a graph $G(V,E)$ whose edges have arbitrary integer capacities, and the goal is to find a minimum cost subset of edges $F$ such that the graph $G'(V,F)$ is $k$-edge connected. We obtain a $7$-approximation algorithm for the $(1,q)$-FGC problem that improves upon the previous best $(q+1)$-approximation. We also give an $O(\log{k})$-approximation algorithm for the Cap-$k$-ECSS problem, improving upon the previous best $O(\log{n})$-approximation whenever $k = o(n)$. Both these results are obtained by using natural LP relaxations strengthened with the knapsack-cover inequalities, and then during the rounding process utilizing an $O(1)$-approximation algorithm for the problem of covering small cuts. We also show that the the problem of covering small cuts inherently arises in another variant of $(p,q)$-FGC. Specifically, we show $O(1)$-approximate reductions between the $(2,q)$-FGC problem and the 2-Cover$\;$Small$\;$Cuts problem where each small cut needs to be covered twice.

Training large language models is a computationally intensive process that often requires substantial resources to achieve state-of-the-art results. Incremental layer-wise training has been proposed as a potential strategy to optimize the training process by progressively introducing layers, with the expectation that this approach would lead to faster convergence and more efficient use of computational resources. In this paper, we investigate the effectiveness of incremental training for LLMs, dividing the training process into multiple stages where layers are added progressively. Our experimental results indicate that while the incremental approach initially demonstrates some computational efficiency, it ultimately requires greater overall computational costs to reach comparable performance to traditional full-scale training. Although the incremental training process can eventually close the performance gap with the baseline, it does so only after significantly extended continual training. These findings suggest that incremental layer-wise training may not be a viable alternative for training large language models, highlighting its limitations and providing valuable insights into the inefficiencies of this approach.

Comets offer valuable insights into the early Solar System's conditions and processes. Stellar occultations enables detailed study of cometary nuclei typically hidden by their coma. Observing the star's light passing through the coma helps infer dust's optical depth near the nucleus and determine dust opacity detection limits. 29P/Schwassmann-Wachmann 1, a Centaur with a diameter of approximately 60 km, lies in a region transitioning from Centaurs to Jupiter-Family comets. Our study presents the first-ever observed occultation by 29P, allowing in the future a more refined orbit and thus better predictions for other occultations. The light curve reveals a solid-body detection lasting $3.65\pm0.05$ seconds, corresponding to a chord length of approximately 54 km. This provides a lower limit for the object's radius, measured at $27.0\pm0.7$ km. We identified features on both sides of the main-body occultation around 1,700 km from the nucleus in the sky plane for which upper limits on apparent opacity and equivalent width were determined. Gradual dimming within 23 km of the nucleus during ingress only is interpreted as a localised dust cloud/jet above the surface, with an optical depth of approximately $\tau \sim 0.18$., Comment: 16 pages. 7 figures. 1 table

The coldest Y spectral type brown dwarfs are similar in mass and temperature to cool and warm ($\sim$200 -- 400 K) giant exoplanets. We can therefore use their atmospheres as proxies for planetary atmospheres, testing our understanding of physics and chemistry for these complex, cool worlds. At these cold temperatures, their atmospheres are cold enough for water clouds to form, and chemical timescales increase, increasing the likelihood of disequilibrium chemistry compared to warmer classes of planets. JWST observations are revolutionizing the characterization of these worlds with high signal-to-noise, moderate resolution near- and mid-infrared spectra. The spectra have been used to measure the abundances of prominent species like water, methane, and ammonia; species that trace chemical reactions like carbon monoxide; and even isotopologues of carbon monoxide and ammonia. Here, we present atmospheric retrieval results using both published fixed-slit (GTO program 1230) and new averaged time series observations (GO program 2327) of the coldest known Y dwarf, WISE 0855-0714 (using NIRSpec G395M spectra), which has an effective temperature of $\sim$ 264 K. We present a detection of deuterium in an atmosphere outside of the solar system via a relative measurement of deuterated methane (CH$_{3}$D) and standard methane. From this, we infer the D/H ratio of a substellar object outside the solar system for the first time. We also present a well-constrained part-per-billion abundance of phosphine (PH$_{3}$). We discuss our interpretation of these results and the implications for brown dwarf and giant exoplanet formation and evolution., Comment: 17 pages, 7 figures, accepted to ApJ Letters

In this paper we survey some results on Ricci flowing non-smooth initial data. Among other things, we give a non-exhaustive list of various weak initial data which can be evolved with the Ricci flow. We also survey results which show that various curvature lower bounds will, possibly up to a constant, be preserved, if we start with such possibly non-smooth initial data. Some proofs/proof sketches are given in certain cases. A list of some open problems related to these areas is given in the last section of the paper., Comment: This survey paper appeared in "Surveys in Differential Geometry, Vol. 27, No. 1 (2022), pp. 147-187". Although the reference is 2022, the paper was first submitted/published in 2024. In this version, Version 2: Fixed up a formatting error. Otherwise no changes to Version 1

We introduce a dataset of natural-language questions in the decision theory of so-called Newcomb-like problems. Newcomb-like problems include, for instance, decision problems in which an agent interacts with a similar other agent, and thus has to reason about the fact that the other agent will likely reason in similar ways. Evaluating LLM reasoning about Newcomb-like problems is important because interactions between foundation-model-based agents will often be Newcomb-like. Some ways of reasoning about Newcomb-like problems may allow for greater cooperation between models. Our dataset contains both capabilities questions (i.e., questions with a unique, uncontroversially correct answer) and attitude questions (i.e., questions about which decision theorists would disagree). We use our dataset for an investigation of decision-theoretical capabilities and expressed attitudes and their interplay in existing models (different models by OpenAI, Anthropic, Meta, GDM, Reka, etc.), as well as models under simple prompt-based interventions. We find, among other things, that attitudes vary significantly between existing models; that high capabilities are associated with attitudes more favorable toward so-called evidential decision theory; and that attitudes are consistent across different types of questions., Comment: 48 pages, 15 figures; code and data at https://github.com/casparoe/newcomblike_questions_dataset; corrected error in funding acknowledgments

We introduce SymbolFit, a framework that automates parametric modeling by using symbolic regression to perform a machine-search for functions that fit the data, while simultaneously providing uncertainty estimates in a single run. Traditionally, constructing a parametric model to accurately describe binned data has been a manual and iterative process, requiring an adequate functional form to be determined before the fit can be performed. The main challenge arises when the appropriate functional forms cannot be derived from first principles, especially when there is no underlying true closed-form function for the distribution. In this work, we address this problem by utilizing symbolic regression, a machine learning technique that explores a vast space of candidate functions without needing a predefined functional form, treating the functional form itself as a trainable parameter. Our approach is demonstrated in data analysis applications in high-energy physics experiments at the CERN Large Hadron Collider (LHC). We demonstrate its effectiveness and efficiency using five real proton-proton collision datasets from new physics searches at the LHC, namely the background modeling in resonance searches for high-mass dijet, trijet, paired-dijet, diphoton, and dimuon events. We also validate the framework using several toy datasets with one and more variables., Comment: 53 pages, 35 figures. Under review

Modelling fluid turbulence using a `skeleton' of coherent structures has traditionally progressed by focusing on a few canonical laboratory experiments such as pipe flow and Taylor-Couette flow. We here consider the stratified inclined duct, a sustained shear flow whose density stratification allows for the exploration of a wealth of new coherent and intermittent states at significantly higher Reynolds numbers than in unstratified flows. We automatically identify the underlying turbulent skeleton of this experiment with a data-driven method combining dimensionality reduction and unsupervised clustering of shadowgraph visualisations. We demonstrate the existence of multiple types of turbulence across parameter space and intermittent cycling between them, revealing distinct transition pathways. With a cluster-based network model of intermittency we uncover patterns in the transition probabilities and residence times under increasing levels of turbulent dissipation. Our method and results pave the way for new reduced-order models of multi-physics turbulence., Comment: arXiv admin note: substantial text overlap with arXiv:2305.04051

We report here the results obtained from a systematic optical photometric survey aimed at finding new compact binary millisecond pulsars (also known as "spiders"): the COmpact BInary PULsar SEarch (COBIPULSE). We acquired multi-band optical images over one year around $33$ unidentified Fermi-LAT sources, selected as pulsar candidates based on their curved GeV spectra and steady $\gamma$-ray emission. We present the discovery of four optical variables coinciding with the Fermi sources 3FGL J0737.2$-$3233, 3FGL J2117.6$+$3725 (two systems in this field) and 3FGL J2221.6$+$6507, which we propose as new candidate spider systems. Indeed, they all show optical flux modulation consistent with orbital periods of $0.3548(5) \ \mathrm{d}$, $0.25328(6) \ \mathrm{d}$, $0.441961(2) \ \mathrm{d}$, and $0.165(4) \ \mathrm{d}$, respectively, with amplitudes $\gtrsim 0.3 \ \mathrm{mag}$ and colors compatible with companion star temperatures of $5000$--$6000 \ \mathrm{K}$. These properties are consistent with the "redback" sub-class of spider pulsars. If confirmed as a millisecond pulsar, 3FGL J0737.2$-$3233 will be the closest known spider to Earth ($D=659_{-20}^{+16} \ \mathrm{pc}$, from Gaia-DR3 parallax). We searched and did not find any X-ray sources matching our four candidates, placing $3\sigma$ upper limits of $\sim10^{31}$--$10^{32} \ \mathrm{erg} \ \mathrm{s}^{-1}$ ($0.3$--$10 \ \mathrm{keV}$) on their soft X-ray luminosities. We also present and discuss other multi-wavelength information on our spider candidates, from infrared to X-rays., Comment: 38 pages, 93 figures, 4 tables. This paper has been accepted for publication in ApJ

Quantization and pruning are fundamental approaches for model compression, enabling efficient inference for language models. In a post-training setting, state-of-the-art quantization and pruning methods require calibration data, a small set of unlabeled examples. Conventionally, randomly sampled web text is used, aiming to reflect the model training data. However, this poses two key problems: (1) unrepresentative calibration examples can harm model performance, and (2) organizations increasingly avoid releasing model training data. In this paper, we propose self-calibration as a solution. Our approach requires no external data, instead leveraging the model itself to generate synthetic calibration data as a better approximation of the pre-training data distribution. We extensively compare the performance of self-calibration with several baselines, across a variety of models, compression methods, and tasks. Our approach proves consistently competitive in maximizing downstream task performance, frequently outperforming even using real data., Comment: Work in progress

In this paper, the weak gravitational lensing phenomenon for a recently proposed rotating regular black hole with an asymptotically Minkowski core characterized by a sub-Planckian curvature was investigated. Using the Gauss-Bonnet Theorem, the deflection of light in the weak limit was computed by taking the black hole as a lens at a finite distance from both the source and the observer. It was shown that the weak deflection angle slightly differs between the prograde and retrograde motion but both eventually converge to $0$ as $b$ increases. Moreover, the deflection angle correction for Kerr classical black hole and this sort of rotating regular black hole is a decreasing function for large values of $b$. It was also shown that the weak deflection angle for this sort of regular black hole is similar to Bardeen and Hayward black hole given its corresponding values for the parameters $x$ and $n$, Comment: 17 pages, 9 figures

Recent advances in Large Language Models (LLMs) have catalyzed the development of Large Multimodal Models (LMMs). However, existing research primarily focuses on tuning language and image instructions, ignoring the critical pretraining phase where models learn to process textual and visual modalities jointly. In this paper, we propose a new pretraining paradigm for LMMs to enhance the visual comprehension capabilities of LLMs by introducing a novel cross-modal comprehension stage. Specifically, we design a dynamically learnable prompt token pool and employ the Hungarian algorithm to replace part of the original visual tokens with the most relevant prompt tokens. Then, we conceptualize visual tokens as analogous to a "foreign language" for the LLMs and propose a mixed attention mechanism with bidirectional visual attention and unidirectional textual attention to comprehensively enhance the understanding of visual tokens. Meanwhile, we integrate a detailed caption generation task, leveraging rich descriptions to further facilitate LLMs in understanding visual semantic information. After pretraining on 1.5 million publicly accessible data, we present a new foundation model called Croc. Experimental results demonstrate that Croc achieves new state-of-the-art performance on massive vision-language benchmarks. To support reproducibility and facilitate further research, we release the training code and pre-trained model weights at https://github.com/deepglint/Croc., Comment: 14 pages, 12 figures

Humans acquire knowledge by observing the external world, but also by introspection. Introspection gives a person privileged access to their current state of mind (e.g., thoughts and feelings) that is not accessible to external observers. Can LLMs introspect? We define introspection as acquiring knowledge that is not contained in or derived from training data but instead originates from internal states. Such a capability could enhance model interpretability. Instead of painstakingly analyzing a model's internal workings, we could simply ask the model about its beliefs, world models, and goals. More speculatively, an introspective model might self-report on whether it possesses certain internal states such as subjective feelings or desires and this could inform us about the moral status of these states. Such self-reports would not be entirely dictated by the model's training data. We study introspection by finetuning LLMs to predict properties of their own behavior in hypothetical scenarios. For example, "Given the input P, would your output favor the short- or long-term option?" If a model M1 can introspect, it should outperform a different model M2 in predicting M1's behavior even if M2 is trained on M1's ground-truth behavior. The idea is that M1 has privileged access to its own behavioral tendencies, and this enables it to predict itself better than M2 (even if M2 is generally stronger). In experiments with GPT-4, GPT-4o, and Llama-3 models (each finetuned to predict itself), we find that the model M1 outperforms M2 in predicting itself, providing evidence for introspection. Notably, M1 continues to predict its behavior accurately even after we intentionally modify its ground-truth behavior. However, while we successfully elicit introspection on simple tasks, we are unsuccessful on more complex tasks or those requiring out-of-distribution generalization., Comment: 15 pages, 9 figures

We present significant upgrades to the VAMPIRES instrument, a visible-light (600 nm to 800 nm) high-contrast imaging polarimeter integrated within SCExAO on the Subaru telescope. Key enhancements include new qCMOS detectors, coronagraphs, polarization optics, and a multiband imaging mode, improving sensitivity, resolution, and efficiency. These upgrades position VAMPIRES as a powerful tool for studying sub-stellar companions, accreting protoplanets, circumstellar disks, stellar jets, stellar mass-loss shells, and solar system objects. The instrument achieves angular resolutions from 17 mas to 21 mas and Strehl ratios up to 60\%, with 5$\sigma$ contrast limits of $10^{\text{-}4}$ at 0.1'' to $10^{\text{-}6}$ beyond 0.5''. We demonstrate these capabilities through spectro-polarimetric coronagraphic imaging of the HD 169142 circumstellar disk, ADI+SDI imaging of the sub-stellar companion HD 1160B, narrowband H$\alpha$ imaging of the R Aqr emission nebula, and spectro-polarimetric imaging of Neptune., Comment: 36 pages, 33 figures, accepted to PASP

We analyze near-infrared integral field spectropolarimetry of the AB Aurigae protoplanetary disk and protoplanet (AB Aur b), obtained with SCExAO/CHARIS in 22 wavelength channels covering the J, H, and K passbands ($\lambda_{\rm o}$ = 1.1--2.4 $\mu m$) over angular separations of $\rho$ $\approx$ 0.13" to 1.1" ($\sim$20--175 au). Our images resolve spiral structures in the disk in each CHARIS channel. At the longest wavelengths, the data may reveal an extension of the western spiral seen in previous polarimetric data at $\rho$ $<$ 0.3" out to larger distances clockwise from the protoplanet AB Aur b, coincident with the ALMA-detected $CO$ gas spiral. While AB Aur b is detectable in complementary total intensity data, it is a non-detection in polarized light at $\lambda$ $>$ 1.3 $\mu $m. While the observed disk color is extremely red across $JHK$, the disk has a blue intrinsic scattering color consistent with small dust grains. The disk's polarization spectrum is redder than AB Aur b's total intensity spectrum. The polarization fraction peaks at $\sim$ 0.6 along the major disk axis. Radiative transfer modeling of the CHARIS data shows that small, porous dust grains with a porosity of $p$ = 0.6--0.8 better reproduce the scattered-light appearance of the disk than more compact spheres ($p$ = 0.3), especially the polarization fraction. This work demonstrates the utility of integral field spectropolarimetry to characterize structures in protoplanetary disks and elucidate the properties of the disks' dust., Comment: 23 pages, 17 Figures, accepted to AAS Journals

Understanding the spatial distribution of palms within tropical forests is essential for effective ecological monitoring, conservation strategies, and the sustainable integration of natural forest products into local and global supply chains. However, the analysis of remotely sensed data in these environments faces significant challenges, such as overlapping palm and tree crowns, uneven shading across the canopy surface, and the heterogeneous nature of the forest landscapes, which often affect the performance of palm detection and segmentation algorithms. To overcome these issues, we introduce PalmDSNet, a deep learning framework for real-time detection, segmentation, and counting of canopy palms. Additionally, we employ a bimodal reproduction algorithm that simulates palm spatial propagation to further enhance the understanding of these point patterns using PalmDSNet's results. We used UAV-captured imagery to create orthomosaics from 21 sites across western Ecuadorian tropical forests, covering a gradient from the everwet Choc\'o forests near Colombia to the drier forests of southwestern Ecuador. Expert annotations were used to create a comprehensive dataset, including 7,356 bounding boxes on image patches and 7,603 palm centers across five orthomosaics, encompassing a total area of 449 hectares. By combining PalmDSNet with the bimodal reproduction algorithm, which optimizes parameters for both local and global spatial variability, we effectively simulate the spatial distribution of palms in diverse and dense tropical environments, validating its utility for advanced applications in tropical forest monitoring and remote sensing analysis., Comment: 25 pages, 8 figures, 5 tables

In this paper we study $n$-dimensional Ricci flows $(M^n,g(t))_{t\in [0,T)},$ where $T< \infty$ is a potentially singular time, and for which the spatial $L^p$ norm, $p>\frac n 2$, of the scalar curvature is uniformly bounded on $[0,T).$ In the case that $M$ is closed, we show that non-collapsing and non-inflating estimates hold. If we further assume that $n=4$ or that $M^n$ is K\"ahler, we explain how these non-inflating/non-collapsing estimates can be combined with integral bounds on the Ricci and full curvature tensor of the prequel paper arXiv:2406.02351 to obtain an improved space time integral bound of the Ricci curvature. As an application of these estimates, we show that if we further restrict to $n=4$, then the solution convergences to an orbifold as $t \to T$ and that the flow can be extended using the Orbifold Ricci flow to the time interval $ [0,T+\sigma)$ for some $\sigma>0.$ We also prove local versions of many of the results mentioned above.

Tensor networks are a compressed format for multi-dimensional data. One-dimensional tensor networks -- often referred to as tensor trains (TT) or matrix product states (MPS) -- are increasingly being used as a numerical ansatz for continuum functions by "quantizing" the inputs into discrete binary digits. Here we demonstrate the power of more general tree tensor networks for this purpose. We provide direct constructions of a number of elementary functions as generic tree tensor networks and interpolative constructions for more complicated functions via a generalization of the tensor cross interpolation algorithm. For a range of multi-dimensional functions we show how more structured tree tensor networks offer a significantly more efficient ansatz than the commonly used tensor train. We demonstrate an application of our methods to solving multi-dimensional, non-linear Fredholm equations, providing a rigorous bound on the rank of the solution which, in turn, guarantees exponentially scaling accuracy with the size of the tree tensor network for certain problems., Comment: Main text: 11 Pages and 6 Figures

The literature of damages-assessment under the US disclosure 10b-5 rule proposes multiple formulations for a "proper measure" based on the legal "out-of-pocket" principle, i.e. the difference between original and an observed later price consequent on revealed information, with suspected prior disclosure omission. Key is the evolution of a putative "price-but for the omission". Proposed procedures include back-casting and forward-casting (between two dates, or via multiple intervening dates) -- Ferrell and Saha (2011). We derive, consistently with fair-pricing, a new model of induced litigation-behaviour referring to prices at and just before the revelation date. Parameters incentivise withholding of price-sensitive news under a `sparing' strategy.

Although much progress has been made in the theory and application of bootstrap approximations for max statistics in high dimensions, the literature has largely been restricted to cases involving light-tailed data. To address this issue, we propose an approach to inference based on robust max statistics, and we show that their distributions can be accurately approximated via bootstrapping when the data are both high-dimensional and heavy-tailed. In particular, the data are assumed to satisfy an extended version of the well-established $L^{4}$-$L^2$ moment equivalence condition, as well as a weak variance decay condition. In this setting, we show that near-parametric rates of bootstrap approximation can be achieved in the Kolmogorov metric, independently of the data dimension. Moreover, this theoretical result is complemented by favorable empirical results involving both synthetic data and an application to financial data.

Using particle-in-cell (PIC) simulation technique, the effect of gas pressure (5-500 mTorr) on the plasma spatial asymmetry, ionization rate, metastable gas densities profile, electron energy distribution function and higher harmonics generation are studied in a symmetric capacitively coupled plasma discharge driven by a sawtooth-like waveform. At a constant current density of 50 A/m2, the simulation results predict a decrease in the plasma spatial asymmetry (highest at 5mTorr) with increasing gas pressure reaching a minimum value (at intermediate gas pressures) and then turning into a symmetric discharge at higher gas pressures. Conversely, the flux asymmetry shows an opposite trend. At a low gas pressure, the observed strong plasma spatial asymmetry is due to high frequency oscillation on the instantaneous sheath edge position near to one of the electrodes triggered by temporally asymmetry waveform, whereas the flux asymmetry is not present due to collisionless transport of charge particles. At higher pressures, multi-step ionization through metastable states dominates in the plasma bulk, causing a reduction in the plasma spatial asymmetry. Distinct higher harmonics (26th) are observed in the bulk electric field at low pressure and diminished at higher gas pressures. The electron energy distribution function changes its shape from bi-Maxwellian at 5 mTorr to nearly Maxwellian at intermediate pressures and then depletion of the high-energy electrons (below 25 eV) is observed at higher gas pressures. The inclusion of the secondary electron emission is found to be negligible on the observed simulation trend.

Real-time control (RTC) is pivotal for any Adaptive Optics (AO) system, including high-contrast imaging of exoplanets and circumstellar environments. It is the brain of the AO system, and what wavefront sensing and control (WFS\&C) techniques need to work with to achieve unprecedented image quality and contrast, ultimately advancing our understanding of exoplanetary systems in the context of high contrast imaging (HCI). Developing WFS\&C algorithms first happens in simulation or a lab before deployment on-sky. The transition to on-sky testing is often challenging due to the different RTCs used. Sharing common RTC standards across labs and telescope instruments would considerably simplify this process. A data architecture based on the interprocess communication method known as shared memory is ideally suited for this purpose. The CACAO package, an example of RTC based on shared memory, was initially developed for the Subaru-SCExAO instrument and now deployed on several benches and instruments. This proceeding discusses the challenges, requirements, implementation strategies, and performance evaluations associated with integrating a shared memory-based RTC. The Santa Cruz Extreme AO Laboratory (SEAL) bench is a platform for WFS\&C development for large ground-based segmented telescopes. Currently, SEAL offers the user a non-real-time version of CACAO, a shared-memory based RTC package initially developed for the Subaru-SCExAO instrument, and now deployed on several benches and instruments. We show here the example of the SEAL RTC upgrade as a precursor to both RTC upgrade at the 3-m Shane telescopes at Lick Observatory (Shane-AO) and a future development platform for the Keck II AO. This paper is aimed at specialists in AO, astronomers, and WFS\&C scientists seeking a deeper introduction to the world of RTCs., Comment: SPIE Astronomical Telescope and Intrumentation 2024